I'm happy I posted this hand 
I really thought I had played it correctly but now I don't really like my call turn. I'd consider calling turn / raising river with a draw if a draw completes river, even if they are not the same.
After the discussion, I'm interested in knowing whether I should bet on the flop.
Preflop
Hero
In this situation I open 40% of the time.
Villain
SB had 3bet 27% SB vs BU (6/22). Let's pretend the sample is sufficient.
I've added some hands to la55i's range and came up with 22+, A2s+, K8s+, Q8s+, J9s+, T8s+, 97s+, 86s+, 75s+, 65s, 54s, A9o+, KTo+, QTo+, JTo (342 combos, 25.79%).
Hero
Given Villain's 3bet SB vs BU stat, I think my 4 bet range is KK+ and my calling range will be wide.
What's the EV of villain's 3bet?
Let's assume he 5bets only KK+, which gives him 50% equity if I 4bet so we can neglect that case.
Let's give him 40% equity postflop if I call (he has initiative and I have position).
Let x be the % of my range that I'm folding.
His preflop EV in BB is EV(fold) + EV(call) = 4.4x - 0.4 * 8.6 * (1-x) = 7.84x - 3.44.
His EV is 0 if I fold ~44% of the time.
Since I open 40% of the hands, I need to call almost 300 combos.
It's probably a bit less because he probably 3bets less than 27% but let's pretend)
I have to keep in mind that this assumes that I have 60% equity postflop, which won't happen if my range is too wide
This range has 282 combos: QQ-66, A2s+, K9s+, Q9s+, J9s+, T8s+, 97s+, 87s, 76s, A9o+, KTo+, QTo+, JTo
That's actually a lot wider than my calling range when I played his hand... (to be fair I didn't have access to his 3bet SB vs BU range at that time)
Flop
Villain
Villain now has 306 combos.
Let's say that he's a decent player and that a third of the time when he has top pair or second pocket pair, he tries to trap me by checking. That's 106 combos and he's going to check / call a third of that, ie 35 combos
He's also going to check / fold garbage (JT J9 etc), which is another 36 combos
And let's says he's going to check/raise his OESD, sets and 2P which represent 13 + 9 + 2 = 24 combos
I think it's a decent strategy (and it's probably overestimating the player, who is not that balanced, but that's a good base).
He's going to cbet 211 combos and check 95. that's 68%, a bit higher than his actual cbet (65%, 15/23). There are adjustments to be made depending on the type of player villain is, but let's use this as a base.
Hero
254 of my combos made it to the flop.
SB is going to cbet 68% of the time, probably around 50% of the pot.
Which means that his flop cbet EV is EV(hero folds) + EV(hero calls / raises) = 19x - 9 (1-x) = 28x - 9, again with x my fold %.
This is EV0 for x ~32%.
SB has checked, as he does 32% of the time.
Let's say I want to bet around 50% of the pot: 9bb in 19.
The EV of my bet is EV(villain folds) + EV(villain calls / raises) = 19x - 9 (1-x) = 28x - 9, again with x my fold equity
But SB will fold 36 of his 95 possible combos, which gives me a fold equity of 38%.
This is enough to make my bet EV+ regardless of my hand.
If villain check/calls with half of his pairs instead of a third, I have only 31% FE, which gives me an EV of -0.32.
Conclusion
All in all, given the ranges I made, I think I can bet all my hands that are not complete trash (32 out of 254 have neither pair nor draw nor overcard, ie 12.6%).
This was pretty interesting to write, congratulations if you have read everything.
I would be very interested in constructive criticism concerning my thinking process 
The end result is a bit unexpected so I may have made incorrect assumptions.
